Hi everybody,

just registered here after beeing stuck for hours. I'm fairly new to OpenGL programming, though I wrote some simple programs and shaders before.

I'm sure somebody around here can tell me, what's going wrong here.

What I'm trying to achieve here is a simple multi-pass shadowmapping rendering:
#1 Render the z-buffer of the light to a texture
#2 Render the shadows to another texture (viewport resolution)
(#3 Do something with this texture, that's not implemented at the moment)
#4 Render the scene and apply the shadows from the previously rendered texture in step #2

For reference, that's what the scene looks like without shadows:


When I display the result of step #2 in the viewport I get, what I expect - the calculated shadows, which are correctly occluded by the geometry:



Now when I simply display the texture rendered in step #2 in the fragment shader of step #4 I get this:



As you can see, this is basically the identical image, but the first and the third cylinder are suddenly missing, eventhough I didn't change anything. So obviously there is something different rendered to the texture then, when I render the same thing to the viewport (maybe some depth/drawing order issue?).


So I guess there must be either something wrong how I setup the FBO or the texture or some settings I set before drawing the scene.

Here I generate the texture used in step #2 and bind it to the FBO:


void CreateRawShadowFBO()
{
GLenum rawShadowsFBOstatus;

// generate and bind the rawShadowTexture
glGenTextures(1, &rawShadowTexture);
glBindTexture(GL_TEXTURE_2D, rawShadowTexture);

// set the parameters
glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_NEAREST);
glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_NEAREST);
glTexParameterf(GL_TEXTURE_2D, GL_TEXTURE_WRAP_S, GL_CLAMP_TO_EDGE );
glTexParameterf(GL_TEXTURE_2D, GL_TEXTURE_WRAP_T, GL_CLAMP_TO_EDGE );

// texture size = viewport size
glTexImage2D( GL_TEXTURE_2D, 0, GL_RGB, (int) observerWidth, (int) observerHeight, 0, GL_RGB, GL_UNSIGNED_BYTE, 0);
glBindTexture(GL_TEXTURE_2D, 0);

// generate and bind the rawShadowsFBO
glGenFramebuffers(1, &rawShadowsFBO);
glBindFramebuffer(GL_DRAW_FRAMEBUFFER, rawShadowsFBO);

// set the rawShadowTexture as render target
glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_TEXTURE_2D, rawShadowTexture, 0);

// check rawShadowsFBO status
rawShadowsFBOstatus = glCheckFramebufferStatusEXT(GL_DRAW_FRAMEBUFFER);
if(rawShadowsFBOstatus != GL_FRAMEBUFFER_COMPLETE_EXT)
printf("GL_FRAMEBUFFER_COMPLETE_EXT failed, CANNOT use rawShadowsFBO\n");

// switch back to window-system-provided framebuffer
glBindFramebufferEXT(GL_DRAW_FRAMEBUFFER, 0);

}

Here's the shadow calculating part (step #2):


void ProcessShadowMap()
{
//set the viewport
glViewport(0, 0, (int) observerWidth, (int)observerHeight);

glClearColor(0,0,0,1);
glClear(GL_COLOR_BUFFER_BIT | GL_DEPTH_BUFFER_BIT);

//set the matrices
M3DMatrix44f shadowMat;
M3DMatrix44f viewMat;
M3DMatrix44f eyeToLightView;

UpdateMatrices(shadowMat, viewMat, eyeToLightView);

glColorMask(GL_TRUE,GL_TRUE, GL_TRUE, GL_FALSE);
glDepthMask(GL_TRUE);
glClear(GL_DEPTH_BUFFER_BIT);

// render the 'raw' shadows to the rawShadowsFBO
glBindFramebuffer(GL_DRAW_FRAMEBUFFER, rawShadowsFBO);

// set the process shadows shader
glUseProgram(procShadowsShader);
// bind the previously rendered depthmap
glBindTexture(GL_TEXTURE_2D, shadowMap);

// set the Model View Projection Matrix for the shader
glUniformMatrix4fv(glGetUniformLocation(procShadowsShader, "mvMatrix"), 1, GL_FALSE, transformPipeline.GetModelViewMatrix());
//Projektionsmatrix
glUniformMatrix4fv(glGetUniformLocation(procShadowsShader, "projMatrix"), 1, GL_FALSE, transformPipeline.GetProjectionMatrix());

// set the ShadowMatrix for the shader
glUniformMatrix4fv(glGetUniformLocation(procShadowsShader, "eyeToLightViewTexMatrix"), 1, GL_FALSE, shadowMat);
glUniformMatrix4fv(glGetUniformLocation(procShadowsShader, "eyeToLightViewMatrix"), 1, GL_FALSE, eyeToLightView);

//calc the light's position, at the origin of the light-view matrix
M3DMatrix44f invLight;
m3dInvertMatrix44(invLight,lightView);
M3DVector4f lp;
m3dLoadVector4(lp,0,0,0,0);
m3dTransformVector4(lightPos, lp, invLight);

// set the light's parameters
glUniform4fv(glGetUniformLocation(procShadowsShader,"lightPos"),1,lightPos);
glUniform1f(glGetUniformLocation(procShadowsShader,"lightNear"),lightNearClip);
glUniform1f(glGetUniformLocation(procShadowsShader,"lightFar"),lightFarClip);
float lightSizeX = (lightSize/lightFrustumWidth)/(float)shadowMapSize;
float lightSizeY = (lightSize/lightFrustumHeight)/(float)shadowMapSize;
glUniform2f(glGetUniformLocation(procShadowsShader,"wLight"),lightSizeX,lightSizeY);
glUniform1i(glGetUniformLocation(procShadowsShader,"shadowMap"),0);

//draw the model
modelBatch.Draw();

//(de)activate stuff
glUseProgram(0);
glBindTexture(GL_TEXTURE_2D, 0);

glBindFramebuffer(GL_DRAW_FRAMEBUFFER, 0);
modelViewMatrix.PopMatrix();
glColorMask(GL_TRUE,GL_TRUE, GL_TRUE, GL_TRUE);
glDepthMask(GL_TRUE);

gltCheckErrors(procShadowsShader);

}


And here's the final step #4, at the moment the shader used here just displays the rendered texture in step #2:


void RenderShadows()
{
//set the viewport
glViewport(0, 0, (int) observerWidth, (int)observerHeight);

glClearColor(0,0,0,1);
glClear(GL_COLOR_BUFFER_BIT | GL_DEPTH_BUFFER_BIT);

//set the projection matrix
viewFrustum.SetPerspective(45.0f,aspect,0.1f,100);
projectionMatrix.LoadMatrix(viewFrustum.GetProjectionMatrix());


//switch cameras
if(!toggleView)
modelViewMatrix.PushMatrix(observerFrame);
else
modelViewMatrix.PushMatrix(lightFrame);


//set the shader program
glUseProgram(displayShadows);
//bind the shadowtexture
glBindTexture(GL_TEXTURE_2D, rawShadowTexture);

// set the Model View Matrix
glUniformMatrix4fv(glGetUniformLocation(displayShadows, "mvMatrix"),
1, GL_FALSE, transformPipeline.GetModelViewMatrix());
//set the Projectionsmatrix
glUniformMatrix4fv(glGetUniformLocation(displayShadows, "projMatrix"),
1, GL_FALSE, transformPipeline.GetProjectionMatrix());

//set the Normalmatrix
glUniformMatrix3fv(glGetUniformLocation(displayShadows, "normalMatrix"),
1, GL_FALSE, transformPipeline.GetNormalMatrix());

//light position
M3DMatrix44f invLight;
m3dInvertMatrix44(invLight,lightView);
M3DVector4f lp;
m3dLoadVector4(lp,0,0,0,0);
m3dTransformVector4(lightPos, lp, invLight);
//light parameters
glUniform4fv(glGetUniformLocation(displayShadows,"lightPos"),1,lightPos);
glUniform4f(glGetUniformLocation(displayShadows,"lightAmbient"),1,1,1,1);
glUniform4f(glGetUniformLocation(displayShadows,"lightDiffuse"),1,1,1,1);
glUniform1f(glGetUniformLocation(displayShadows,"lightNear"),lightNearClip);
glUniform1f(glGetUniformLocation(displayShadows,"lightFar"),lightFarClip);
float lightSizeX = (lightSize/lightFrustumWidth)/(float)shadowMapSize;
float lightSizeY = (lightSize/lightFrustumHeight)/(float)shadowMapSize;
glUniform2f(glGetUniformLocation(displayShadows,"wLight"),lightSizeX,lightSizeY);
glUniform1i(glGetUniformLocation(displayShadows,"shadowMap"),0);


//draw the model
modelBatch.Draw();

glUseProgram(0);
glBindTexture(GL_TEXTURE_2D, 0);

modelViewMatrix.PopMatrix();
gltCheckErrors(displayShadows);
}



Anybode an idea what goes wrong, when rendering to the texture? I'd be really grateful for any hints

You don't have a depth buffer in your FBO, so the geometry just appears in the order it's drawn. You need to have a depth buffer in step 2 too since you're trying to draw all the geometry again, and you obviously need the closest depth there.

I get the feeling you're trying to do a screen space blur of the shadow map to get smooth shadows, correct? This won't look very good and will be very inaccurate, especially for steep angles since it's uniformly blurred. You'll also possible get light bleeding where you're not supposed to unless you take the depth of neighboring pixels into account. Anyway, it's far from optimal. The easiest way to do soft shadows is with PCF, which is basically sampling lots of nearby depth values from the shadow map and checking how many of these that pass. This gives problems with self shadowing though so you need a high depth bias which in turn pretty much removes self shadowing. If you're feeling really adventurous look up variance shadow maps.

Hey agentd,

first of all, thank you a lot for your reply!

I suspected something like that, but didn't know I had to explicitely create a depth buffer FBO for it. So I tried to do my homework and checked several (very similar) tutorials about it, but after implementing the following code my texture just stays black (rest of the code didn't change)

void CreateRawShadowFBO()
{
GLenum rawShadowsFBOstatus;

// generate and bind the rawShadowTexture
glGenTextures(1, &rawShadowTexture);
glBindTexture(GL_TEXTURE_2D, rawShadowTexture);

// set the parameters
glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_NEAREST);
glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_NEAREST);
glTexParameterf(GL_TEXTURE_2D, GL_TEXTURE_WRAP_S, GL_CLAMP_TO_EDGE );
glTexParameterf(GL_TEXTURE_2D, GL_TEXTURE_WRAP_T, GL_CLAMP_TO_EDGE );

// texture size = viewport size
glTexImage2D( GL_TEXTURE_2D, 0, GL_RGB, (int) observerWidth, (int) observerHeight, 0, GL_RGB, GL_UNSIGNED_BYTE, 0);

// unbind texture
glBindTexture(GL_TEXTURE_2D, 0);


// generate and bind depth buffer
glGenRenderbuffers(1, &rawShadowsDepthBuffer);
glBindRenderbuffer(GL_RENDERBUFFER, rawShadowsDepthBuffer);
glRenderbufferStorage(GL_RENDERBUFFER, GL_DEPTH_COMPONENT, (int) observerWidth, (int) observerHeight);
//unbind depth buffer
glBindRenderbuffer(GL_RENDERBUFFER, 0);

// generate and bind the rawShadowsFBO
glGenFramebuffers(1, &rawShadowsFBO);
glBindFramebuffer(GL_FRAMEBUFFER, rawShadowsFBO);

// attach the color texture to the frame buffer
glFramebufferTexture2D(GL_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_TEXTURE_2D, rawShadowTexture, 0);
// attach depth buffer
glFramebufferRenderbuffer(GL_FRAMEBUFFER, GL_DEPTH_ATTACHMENT, GL_RENDERBUFFER, rawShadowsDepthBuffer);

// check rawShadowsFBO status
rawShadowsFBOstatus = glCheckFramebufferStatus(GL_FRAMEBUFFER);
if(rawShadowsFBOstatus != GL_FRAMEBUFFER_COMPLETE)
printf("GL_FRAMEBUFFER_COMPLETE failed, CANNOT use rawShadowsFBO\n");

// switch back to window-system-provided framebuffer
glBindFramebuffer(GL_DRAW_FRAMEBUFFER, 0);
}


If I don't attach the depth buffer FBO ([size=2][font=courier new,courier,monospace]glFramebufferRenderbuffer(GL_FRAMEBUFFER, GL_DEPTH_ATTACHMENT, GL_RENDERBUFFER, rawShadowsDepthBuffer[/font]); ) I get the same result as before. So I'm afraid my depth buffer FBO is setup in a wrong way, though I'm unable to see what's the issue, as I can't see any difference to the 3 tutorials I checked on that matter.

That's the thing I which makes it sometimes hard to enjoy OpenGL coding for a beginner like me, as it's really hard to debug. Feels like a big blackbox.

Your assumption was right. I'm trying to achieve something along that matter, though I don't plan to apply a uniform blur. Had an idea to compensate the viewing angle dependent issue you mentioned. However I've no idea if this will work out as I've planned, but that doesn't really matter, as I'm having fun playing around with stuff like that.

I did implement both PCF and PCSS shadow mapping a while ago (2nd with rotated poisson disc):



So as I mentioned before, my goal here is not really about just implementing a shadow algorithm, but more playing around with my own ideas (as bad as they might turn out ;) ). I also looked into VSM when I did the other shadow mapping techniques a while back, but felt like it had too many limitations.

I think the rendering is working fine, but you're not clearing the FBO, causing the depth test to fail after the first frame. Make sure that all FBOs are cleared properly, both the shadow map and the shadow post processing buffer. I see for example that you call glViewport(), then glClear() and THEN glBindFramebuffer(). I don't think you wanted to clear a shadow map sized part of the screen, right? =S

Of course :facepalm: I cleared before binding the buffer. Just had to swap these in the code and now everything works fine.

Thanks again